MAROKO133 Eksklusif ai: Canadian ground robot survives Atlantic drop, reaches Portuguese s

📌 MAROKO133 Eksklusif ai: Canadian ground robot survives Atlantic drop, reaches Po

Rheinmetall Canada tested a Mission Master 2.0 unmanned ground vehicle by dropping it from a warship into the Atlantic and driving it to a Portuguese beach. 

The trial took place during NATO’s Robotic Experimentation and Prototyping using Unmanned Systems exercise on September 23. The drill gathered 24 countries and evaluated more than 276 unmanned systems for maritime operations.

“[Mission Master 2.0] was embarked on a warship that navigated into challenging waters and was dropped into the sea, with a crane, and its mission was to navigate back to the beach near the testing site effectively,” Rheinmetall Canada’s Director of International Business Development Étienne Rancourt told Defense News. 

The successful run showed the UGV could complete a sea-to-shore transit and then operate on land without direct human towing or recovery.

Modular kit for maritime tasks

The amphibious trial used a Mission Master loaded with a partner kit. That included a tethered drone supplied by France’s Elistair, an Echodyne radar from the United States, and Rheinmetall’s mast and sensor package.

The configuration aimed to support scouting, sensing, and communications during an amphibious approach, roles navies and marine units increasingly want from unmanned systems.

Rancourt said the tested vehicle incorporated recent operator feedback. “The version you see here has upgrades based on current operators’ feedback, including the U.S., Norway, the UK, etc. These are mainly all focused on increased robustness and stability of the vehicle,” he said.

Those upgrades target reliability in saltwater conditions and stability when crossing shoreline terrain.

Mission Master family and autonomy

The defense technology company’s Mission Master line covers several UGV sizes and mission sets. The family handles surveillance, logistics, casualty evacuation, communications relay, and fire support.

All models use Rheinmetall’s Path autonomous kit, or A-kit, which enables follow, convoy, and fully autonomous modes to reduce routine tasks for soldiers.

“Rheinmetall is committed to keeping a human in the loop in all kinetic operations, assuring that it is never a machine that decides when to open fire,” the company stated on its website.

 “The Path A-kit has an open, flexible architecture, meaning it will rapidly integrate first-hand innovations as artificial intelligence technologies evolve.” 

That architecture is intended to let armed forces swap sensors and software without replacing the whole vehicle.

Wider trends and partnered systems

The Portugal exercise highlights a broader interest in unmanned systems that can bridge sea and land. Militaries want robots that scout ahead, gather intelligence on beaches and littoral zones, and carry supplies during complex water-to-shore moves.

Other nations are already fielding amphibious UGVs in combat zones, and multinational exercises are testing which roles make sense for a partnership between surface, aerial, and ground robots.

Rheinmetall also displayed partnered weapons and rapid-response systems during the event. A Hero-120 loitering munition from Uvision, shown alongside the Mission Master, was used in a mock quick reaction scenario. 

“We were shortly notified of our mission, and within two minutes of receiving the coordinates of the target, we were set up on the beach to launch and strike it,” a Uvision USA representative said, according to reports..

That demo underlined how sensor-equipped ground robots can feed targeting and launch chains during near-shore operations.

🔗 Sumber: interestingengineering.com

📌 MAROKO133 Eksklusif ai: MIT Researchers Unveil “SEAL”: A New Step Towards Self-I

The concept of AI self-improvement has been a hot topic in recent research circles, with a flurry of papers emerging and prominent figures like OpenAI CEO Sam Altman weighing in on the future of self-evolving intelligent systems. Now, a new paper from MIT, titled “Self-Adapting Language Models,” introduces SEAL (Self-Adapting LLMs), a novel framework that allows large language models (LLMs) to update their own weights. This development is seen as another significant step towards the realization of truly self-evolving AI.

The research paper, published yesterday, has already ignited considerable discussion, including on Hacker News. SEAL proposes a method where an LLM can generate its own training data through “self-editing” and subsequently update its weights based on new inputs. Crucially, this self-editing process is learned via reinforcement learning, with the reward mechanism tied to the updated model’s downstream performance.

The timing of this paper is particularly notable given the recent surge in interest surrounding AI self-evolution. Earlier this month, several other research efforts garnered attention, including Sakana AI and the University of British Columbia’s “Darwin-Gödel Machine (DGM),” CMU’s “Self-Rewarding Training (SRT),” Shanghai Jiao Tong University’s “MM-UPT” framework for continuous self-improvement in multimodal large models, and the “UI-Genie” self-improvement framework from The Chinese University of Hong Kong in collaboration with vivo.

Adding to the buzz, OpenAI CEO Sam Altman recently shared his vision of a future with self-improving AI and robots in his blog post, “The Gentle Singularity.” He posited that while the initial millions of humanoid robots would need traditional manufacturing, they would then be able to “operate the entire supply chain to build more robots, which can in turn build more chip fabrication facilities, data centers, and so on.” This was quickly followed by a tweet from @VraserX, claiming an OpenAI insider revealed the company was already running recursively self-improving AI internally, a claim that sparked widespread debate about its veracity.

Regardless of the specifics of internal OpenAI developments, the MIT paper on SEAL provides concrete evidence of AI’s progression towards self-evolution.

Understanding SEAL: Self-Adapting Language Models

The core idea behind SEAL is to enable language models to improve themselves when encountering new data by generating their own synthetic data and optimizing their parameters through self-editing. The model’s training objective is to directly generate these self-edits (SEs) using data provided within the model’s context.

The generation of these self-edits is learned through reinforcement learning. The model is rewarded when the generated self-edits, once applied, lead to improved performance on the target task. Therefore, SEAL can be conceptualized as an algorithm with two nested loops: an outer reinforcement learning (RL) loop that optimizes the generation of self-edits, and an inner update loop that uses the generated self-edits to update the model via gradient descent.

This method can be viewed as an instance of meta-learning, where the focus is on how to generate effective self-edits in a meta-learning fashion.

A General Framework

SEAL operates on a single task instance (C,τ), where C is context information relevant to the task, and τ defines the downstream evaluation for assessing the model’s adaptation. For example, in a knowledge integration task, C might be a passage to be integrated into the model’s internal knowledge, and τ a set of questions about that passage.

Given C, the model generates a self-edit SE, which then updates its parameters through supervised fine-tuning: θ′←SFT(θ,SE). Reinforcement learning is used to optimize this self-edit generation: the model performs an action (generates SE), receives a reward r based on LMθ′’s performance on τ, and updates its policy to maximize the expected reward.

The researchers found that traditional online policy methods like GRPO and PPO led to unstable training. They ultimately opted for ReST^EM, a simpler, filtering-based behavioral cloning approach from a DeepMind paper. This method can be viewed as an Expectation-Maximization (EM) process, where the E-step samples candidate outputs from the current model policy, and the M-step reinforces only those samples that yield a positive reward through supervised fine-tuning.

The paper also notes that while the current implementation uses a single model to generate and learn from self-edits, these roles could be separated in a “teacher-student” setup.

Instantiating SEAL in Specific Domains

The MIT team instantiated SEAL in two specific domains: knowledge integration and few-shot learning.

• Knowledge Integration: The goal here is to effectively integrate information from articles into the model’s weights.
• Few-Shot Learning: This involves the model adapting to new tasks with very few examples.

Experimental Results

The experimental results for both few-shot learning and knowledge integration demonstrate the effectiveness of the SEAL framework.

In few-shot learning, using a Llama-3.2-1B-Instruct model, SEAL significantly improved adaptation success rates, achieving 72.5% compared to 20% for models using basic self-edits without RL training, and 0% without adaptation. While still below “Oracle TTT” (an idealized baseline), this indicates substantial progress.

For knowledge integration, using a larger Qwen2.5-7B model to integrate new facts from SQuAD articles, SEAL consistently outperformed baseline methods. Training with synthetically generated data from the base Qwen-2.5-7B model already showed notable improvements, and subsequent reinforcement learning further boosted performance. The accuracy also showed rapid improvement over external RL iterations, often surpassing setups using GPT-4.1 generated data within just two iterations.

Qualitative examples from the paper illustrate how reinforcement learning leads to the generation of more detailed self-edits, resulting in improved performance.

While promising, the researchers also acknowledge some limitations of the SEAL framework, including aspects related to catastrophic forgetting, computational overhead, and context-dependent evaluation. These are discussed in detail in the original paper.

Original Paper: https://arxiv.org/pdf/2506.10943

Project Site: https://jyopari.github.io/posts/seal

Github Repo: https://github.com/Continual-Intelligence/SEAL

The post MIT Researchers Unveil “SEAL”: A New Step Towards Self-Improving AI first appeared on Synced.

🔗 Sumber: syncedreview.com